On Feb 3, 2016, at 1:07 PM, Darrick J. Wong wrote: > > On Wed, Feb 03, 2016 at 11:38:33AM -0800, Joe Perches wrote: >> On Wed, 2016-02-03 at 11:32 -0800, Darrick J. Wong wrote: >>> On Wed, Feb 03, 2016 at 10:30:32AM -0800, Joe Perches wrote: >>>> On Wed, 2016-02-03 at 10:13 -0800, Darrick J. Wong wrote: >>>>> On Wed, Feb 03, 2016 at 09:19:06PM +0800, Huaitong Han wrote: >>>>>> This patch adds a line break for proc mb_groups display. >>>> >>>> Using 2 lines for output might break any existing users. >>>> >>>> Are there any? >>> >>> It's a multiline file if you have more than one blockgroup; this just makes it >>> so that you don't have to special-case BG 0. >> >> And existing scripts might do that now and might fail >> to do properly after this change. > > Or they might have sed -e 's/]#0/]\n#0/g' in which case they won't be affected. I suspect that any scripts which handled this in the past probably didn't even notice and just missed the bg=0 line at the end of the header. Users looking at the output probably saw it line-wrapped by the terminal and didn't notice either. >>> IOW: mb_groups scripts already had to parse multiple lines, and most likely any >>> script parsing it would inject a newline after the header. >> >> I've no dog in this fight really. I just wanted to make >> it clear that this could cause existing scripts to fail. >> >> proc output is supposed to be unchanging except maybe >> adding new fields to existing lines. >> >> Your choice. > > Ted's, really. I have no idea which scripts do with various per-fs /proc > files. Usually poking in mb_groups is only done as part of failure report data > collection to see what's mucked up the fs this time. > > Anyway, I'll defer to the maintainer. :) I think it makes sense to accept the patch, since I doubt any scripts will be broken, and it is the "right thing to do" rather than perpetuate a bug. Cheers, Andreas